Abstract The study aimed to authenticate milk samples derived from maize silage (MS)-, crop silage/hay (SH)- and grassland hay (GH)-based diets through direct analysis in real time coupled to high resolution mass spectrometry (DART-HRMS). A mid-level data fusion partial least square–discriminant analysis (PLS-DA) allowed the correct separation of milk samples, as confirmed by a linear discriminant analysis (LDA) validation procedure. A pool of ions (absolute loading value > 0.3) of a set of four (two extraction solvents per two ion modes) PLS-DA was selected, fused and submitted to hierarchical cluster analysis. This approach identified the 25 most informative DART-HRMS biomarkers: lactate, glutamate and hydroxycinnamic acid for MS; creatinine, methyl 2-furoate, norgramine, C18:2, C20:2 and C22:2 monoacylglycerols for SH; palmitate, O-methylated flavonoid and glyceric acid for GH. This restricted (+/−) DART-HRMS pool of low-molecular-weight biomolecules confirmed that forage botanical origin and conservation method affect the milk metabolomics profile.

中文翻译：

---

通过 (+/-) DART-HRMS 签名的中级数据融合对基于草料的牛奶进行认证

摘要 本研究旨在通过实时直接分析结合高分辨率质谱 (DART-HRMS) 对源自玉米青贮 (MS)、作物青贮/干草 (SH) 和草原干草 (GH) 的饲料的牛奶样品进行鉴定。 ）。中级数据融合偏最小二乘判别分析 (PLS-DA) 允许正确分离牛奶样品，正如线性判别分析 (LDA) 验证程序所证实的那样。选择一组四组（每两种离子模式两种提取溶剂）PLS-DA 的离子池（绝对负载值 > 0.3），融合并提交给分层聚类分析。这种方法确定了 25 个信息量最大的 DART-HRMS 生物标志物：用于 MS 的乳酸、谷氨酸和羟基肉桂酸；肌酐、2-糠酸甲酯、诺格明、用于 SH 的 C18:2、C20:2 和 C22:2 单酰基甘油；棕榈酸酯，用于 GH 的 O-甲基化黄酮类化合物和甘油酸。这种受限制的 (+/-) DART-HRMS 低分子量生物分子库证实了草料植物来源和保存方法会影响牛奶代谢组学特征。